Upon careful scrutiny, however, it will be found
possible to decide, by examining their melodic bearing, to which phrase
they pertain. In Example 22, they are manifestly (even in No. 5) a
part of the present phrase; in Example 23 and 24 they are as certainly
preliminary to the phrase which follows. In the following example they
seem to constitute an entirely independent little "interlude," without
direct reference to either phrase:

[Illustration: Example 25. Fragment of Mozart.]

* * * * * *

THE ELISION.--Finally, there are some (very rare) instances where the
composer appears to yield to the seductive influence of such extensive
preliminary groups as those seen in Example 24, and by setting aside
the trifling discrepancy, permits the apparent preliminary tones to
represent the _actual first measure of the next phrase_. This is
easily accomplished, when, as in Example 24, No. 2, it is only one
16th-note short of a full measure. And although this 16th, being the
cadence-chord, is actually equivalent to the whole measure, it is
sometimes less confusing to the hearer to silence it. This is called
stifling the cadence (or Elision); and its presence depends simply upon
sufficient proof that what was supposed to be the cadence-measure (and
to a certain extent is such) is at the same time _really the first
measure of the next sentence_. The following contains an illustration
of the elision of a cadence:

[Illustration: Example 26. Fragment of Mozart.]

[Illustration: Example 26 continued.]

The proofs of this very singular and apparently untrustworthy analysis
are: (1) That there is absolutely no doubt about the first cadence,
marked *; (2) that a cadence is consequently due, and expected, four
measures later,--this proving the measure in question to be the
"cadence-measure of the old phrase," as it is marked and as it appeals
to our sense of cadence; (3) that the last four measures unmistakably
represent a regular, compact phrase,--this proving that the
"cadence-measure of the old phrase" is unquestionably _at the same time
the first measure, or actual beginning, of the new phrase_. In a word,
one measure is lost--not in effect, for the elements of the expected
cadence are all present,--but in the counting. This lost measure is
the stifled cadence-measure, omitted by Elision.

Such cases are, as stated, very rare; so rare that the student will do
wisely to leave them quite